April 28, 2026
There’s a lot of AI doomerism. Memes about having a short window to “escape the permanent underclass” or be crushed by a wave of AI automation. Unparalleled levels of FOMO reading about how everyone is vibe coding entire businesses overnight. A few vague but plausible-sounding headlines about a bad job market for new grads (the Times publishes a variation of this article every year, even the good ones).
The scariest headlines are about layoffs. They range from out of touch to self-serving to apocalyptic (or all three).
When Block announced a 40% cut, Jack Dorsey wrote one of the best statements you’ll see from a public company. It read as sincere and heartfelt.
But it wasn’t entirely honest. Part of Block’s reasoning was the use of “intelligence tools.” The truth is that Block, like many tech companies, massively over-hired during Covid. It wasn’t a bad idea at the time, it just turned out to be wrong.
Block is not the only company guilty of this. Every big tech company (except Apple) followed a similar trajectory. They incorrectly extrapolated the explosive growth from the pandemic and were not prepared for the sudden end of a decade of ZIRP.
Public companies do this all the time because over-hiring is not a fatal mistake. But failing to capitalize on a new opportunity is.
Layoffs are quickly forgotten and blue-chip companies like Google and Meta retain their strong reputations for recruiting new talent.
Some of these aren’t even real “layoffs”—they’re just business as usual, cutting bottom performers and business units that are no longer a priority. Meta does this aggressively. Companies outside of tech do it all the time too. Some hires work out, some don’t.
Big Tech new-hire interviews yield a lot of false positives so cutting low performers is an important part of the process. Engineering interviews do not evaluate a candidate’s ability to ship working software or be an effective teammate. They only test the ability to grind LeetCode style questions. The best teammates I’ve ever had might ace or bomb this interview on any given day. And some of the most useless teammates are the most brilliant LeetCode problem solvers.
All this means big tech makes a lot of hiring mistakes. Layoffs have nothing to do with being replaced by AI. The press release will always reflect the best narrative and claiming AI efficiency is a better story for a public company than just admitting they over-hired.
One recent exception: Meta came close to telling the truth when they announced an upcoming 10% reduction. They said the move will “offset the other investments we’re making,” presumably in AI. That’s probably true. Their stock was crushed when Zuck was on pace to plow $250b, the equivalent of the Apollo space program, into the failed Metaverse effort. If that investment was offset by cost-cutting maybe Wall Street would have been more accepting.
There are other reasons to be skeptical of the AI job loss narrative.
First, it’s just not good enough yet to do any of the jobs that are being eliminated. (The only VC that will actually say this out loud is Marc Andreesen.)
Second, AI makes it feasible to tackle projects that previously weren’t worthwhile. So there is now more work, not less. Jevons Paradox is frequently cited as a reason that AI use will continue to grow but it also applies to any work that can be made more efficient by AI use.
Third, employees that can leverage AI to be more productive are more valuable than before. Why would an employer want to get rid of somebody that is now delivering more?
It’s crazy to me that the consensus right now is that AI inevitably leads to job losses. You can already see evidence that this isn’t actually true for software, the field most impacted by generative AI.